How to implement role based security for SQL queries

Description

The Atlassian folks come up with a way to implement role-based security based on user ids to help them manage their data access better. This requires add-on version 5.2 and is simple to configure for the add-on using Data Source Profiles. This does require your database be set up so that roles are defined for each user in Confluence.

Steps

Have the database administrator define roles for each Confluence user by user id and set appropriate authority to data based on roles. A role for anonymous should also be added.
Go to UPM and find the SQL for Confluence
Press the Configure button
Add a data source profile
- Add a dataSource parameter to reference an application server data source if it has a different name that the profile
- Add beforeSql and afterSql parameters similar to the example below
The new profile is available immediately - test it on a page to ensure expected behavior

Implementation without page updates

Once you are finished testing, you can use data source alias support to change an existing (application server) data source to use the new role-based security model without requiring any page updates. Simply rename the data source profile to the same name as the existing (application server) data source already used in the SQL macros.

Cache macro

The Cache Macro should not be used around any content that is rendered differently for different users. The role-based model described here renders differently depending on the user, so the Cache Macro should not be used in this case.

Example

For Postgres, the following SQL works:

beforeSql=begin; SET LOCAL ROLE @user_id@; 
afterSql=; rollback;

The substitution variable @user_id@ is replaced in SQL by the SQL macros with the current logged in user id or anonymous (if not logged in).